Análise dos movimentos conjuntos entre retornos de ativos brasileiros de renda variável e de renda fixa e índices estrangeiros no período 2004-2012

• Ano de defesa: 2013
• Autor(a) principal: Getulio Alves de Souza Matos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Vetor autorregressivo; Comovimentos; Vetor de correção de erros; Causalidade de Granger; GARCH multivariado; Renda fixa; Renda variável; Índices; Títulos (Finanças); Finanças Brasil; Administração
• Link de acesso: http://hdl.handle.net/1843/BUOS-96UJXB

With the increasing expansion of financial products and services, Brazilian market offers to investors, domestic or foreign, several alternatives for the allocation of resources. However, the apparent diversification of investments cannot be well performed, since the comovements in prices or returns may reveal a false or naive diversification. In this context, this dissertation analyzes contemporary movements among returns of different investment alternatives offered in Brazilian market, different indices of fixed income and stock markets, due to variations of foreign benchmarks (U.S. Dollar and the Dow Jones Indexes Industrial Average and UBS Commodities), gold and Risk Brazil (Emerging Markets Bond Index - Brazil). With a view to better understanding of the structure of the domestic financial market, the econometric modeling of the data comprehended Causality relations (Granger), and the estimations of a Vector Autoregressive (VAR) and of a multivariate GARCH model for return series. The results of the three techniques applied to return series converge to reveal relations among the indexes under analysis, revealing the lead-and-lag relations between: the lagged returns of EMBI+Brazil and the returns of Ibovespa; the lagged and current returns of IMA; and the lagged returns of Ibovespa and gold for the returns of the exchange rates. In addition to these findings, the MGARCH model revealed the Dow Jones as explanatory variable to the exchange rate. The analysis was also splitted in two subsamples, divided at the date of Lehman Brothers bankruptcy. This procedure revealed a few differences in the explanatory variables for each one of the smaller samples. Alternatively, cointegration relations were tested for log-prices series, also splitted in three subsamples. Via a Vector Error Correction Model (VEC), no cointegrating vectors were found in the subsamples. As a remark, there are possible improvements to be done, by developing new indices for Brazilian market, and the need for analysis of macroeconomic factors coupled to the dynamics between the sectors of the domestic market to estimate consistent results on the long term, even in turbulent times.